Inferring Single-Cell 3D Chromosomal Structures Based on the Lennard-Jones Potential

Area of Science:

• Genomics
• Computational Biology
• Biophysics

Background:

• Reconstructing three-dimensional (3D) chromosomal structures from single-cell Hi-C data is challenging due to data sparseness.
• Accurate 3D genome organization is crucial for understanding gene regulation and cellular function.

Purpose of the Study:

• To develop a robust computational method for reconstructing 3D chromosomal structures from sparse single-cell Hi-C data.
• To model chromosomal structures at both 500 kb and 50 kb resolutions.

Main Methods:

• Utilized the Lennard-Jones potential to represent DNA as beads in a 3D lattice.
• Employed a 2D Gaussian function for imputing sparse Hi-C contact matrices.
• Designed a novel loss function incorporating Lennard-Jones potential (ε value) for binding stability.
• Applied Metropolis-Hastings algorithm and simulated annealing for structural optimization.

Main Results:

• Successfully reconstructed 3D chromosomal structures at 500 kb and 50 kb resolutions.
• The novel loss function effectively guided the optimization process based on Hi-C contact data.
• Validated the accuracy and correctness of the reconstructed models using multiple criteria and 3D-FISH data.

Conclusions:

• The Lennard-Jones potential-based approach provides a valid and effective method for 3D genome structure reconstruction from sparse single-cell Hi-C data.
• This method offers a significant advancement in modeling high-resolution 3D chromosomal organization.
• The validated models contribute to a better understanding of genome architecture and its functional implications.